The invention relates to a method of detecting a routing loop in a telecommunication network equipped with nodes, each containing node-specific routing information, in which a path is assembled from a source node, on the one hand, to a destination node, on the other hand, on the basis of the node-specific routing information associated with the source node and the node-specific routing information associated with the nodes situated on the path.
Such a method is disclosed by U.S. Pat. No. 5,014,262. In the latter, two submethods are described: a first submethod for detecting a routing loop (call looping) and a second submethod for detecting inefficient routing (circuitous routing). The first submethod, which is carried out for each destination node, comprises four steps. In the first step, all the nodes which do not receive any traffic for the destination node are removed, and in the second step, all the nodes which do not transmit any traffic to the destination node are removed. These nodes which have to be removed can, after all, not form part of a routing loop. In the third step, all tee nodes which form part of so-called miniloops are dealt with. Two nodes form part of a miniloop if one of the nodes receives traffic for the destination node from the other node and the first node also transmits traffic to the other node. For technical reasons, a miniloop can never occur as a routing loop because, at the position of a node, which preceding node the traffic to be routed further originates from is known and said traffic will never, of course, need to be sent back to said previous node. In the fourth step, the nodes which do not form part of a routing loop are identified. Such a node must, for example, meet the criterion that, if the node transmits traffic to another node, said other node must not send any traffic back to the node concerned. How this criterion, which in fact only states in other words that no routing loop must be possible, must be met is unclear. One solution could be to investigate for all the nodes left over after the first three steps whether the node which has the most incoming and outgoing traffic and transmits traffic to particular nodes receives traffic back from said particular nodes via a longer path. The said node would then have to be removed and the remaining nodes would have to be investigated again, etc. Like the criterion stated, such investigations are, however, yet another formulation of the problem of detecting routing loops. This known method, which is in any case cumbersome and, consequently, time-consuming, furthermore has the disadvantage that it is uncertain whether existing routing loops are found with it.